Improving Search Suggestions Using SOLR

What are some possible ways you could improve forward-type sentences (auto-complete) that appear in free-form searches?

In my opinion, all suggestions that appear for keywords are stored in the SOLR table.

How do you ensure that it covers all industry-specific, relevant forward-looking offers?

Can you automate, including the most recent user queries, which currently do not produce search results, to result in relevant ones?

+4
source share
1 answer

During pre-processing, documents uploaded to the search engine should be enriched with what is reasonable and helps to find them. For instance. a document containing the string paris can be enriched with french capital , capital of france , ile-de-france , ... For this you need a dictionary. You can take data from dbpedia.org or only in English - WordNet. In order not to overly generalize, you will need to implement some ambiguity (meaning opening) in the first stage, since paris - for example - can be divided into alexandros , alaksandu of wilusa , king of troy , depending on the context.

0
source

Source: https://habr.com/ru/post/1440070/


All Articles